Ctrl K

Quick Ref / Command

Engineer Kit

/ak:scenario

Edge-case test scenarios

Expand a feature or code path into edge cases and test scenarios across 12 dimensions, with one-shot or iterative saturation modes.

Read target

Filter dimensions

Generate scenarios

Classify severity

Rule 01

A skipped dimension needs a stated assumption

Rule 02

An unsafe skip becomes its own scenario

Rule 03

Iterative mode values novelty over raw volume

Rule 04

Critical and High rows should feed test planning or implementation risk

Execution Map

Core lanes from input to output, aligned with the command's real execution path.

Start

Input, scope, route

1 Read target Read the file path or parse the feature description to identify actors, components, preconditions, and scope.
2 Filter dimensions Decide which dimensions apply; explicitly skip irrelevant ones and state the assumption behind each skip.

Work

Agent / skill execution

3 Generate scenarios In one-shot mode, create 3–5 scenarios per relevant dimension with expected behavior.
4 Classify severity Rate Critical, High, Medium, or Low based on data loss, security, broken subsets, recoverable UX, or minor glitches.

Verify

Gate, review, validation

5 Iterate when requested With `--iterations N` or `--saturation`, loop through dimensions, combinations, negations, personas, and temporal shifts while logging novelty.
6 Track novelty Keep new and useful variants, discard duplicates or out-of-scope rows, rotate dimensions after three same-dimension iterations.

Report, handoff, artifact

7 Output coverage Return a scenario report with analyzed/skipped dimensions, table, severity totals, and coverage matrix for iterative runs.

Syntax, arguments, subcommands

How to invoke the skill: syntax, positional arguments, shared options, then each subcommand with its own syntax and outcome.

Syntax

/ak:scenario <file path or feature description> [--iterations N] [--saturation] [--domain <type>] [--focus <dim>] [--format <type>]

Arguments

<file path or feature description> Required

Target to explore

A narrow code path, file path, or natural-language feature description that defines the behavior and evidence scope. Include actors, state, data, integrations, constraints, and known failure modes when they are not obvious from the file.

/ak:scenario src/api/account-recovery.ts

Shared options

--iterations N

Fixed iteration count

Run exactly N scenario iterations, then summarize. This is a budget and hard ceiling, not proof of exhaustive coverage.

/ak:scenario src/api/payment.ts --iterations 25

--saturation

Novelty stop

Continue until two consecutive iterations produce no New classification. It has no separate numeric maximum in the shipped contract.

/ak:scenario "Add multi-tenancy to the database layer" --saturation

--domain <type>

Domain hint

Prioritize software, product, business, security, or marketing context. It guides ordering and emphasis; it does not narrow the evidence scope by itself.

/ak:scenario src/middleware/auth.ts --saturation --domain security

--focus <dim>

Focus dimension

Prioritize edge-cases, failures, security, or scale during exploration while still reporting analyzed and skipped dimensions.

/ak:scenario "Account recovery" --focus failures

--format <type>

Report shape

Select table, use-cases, test-scenarios, or threat-scenarios output. The option changes report shape, not whether scenarios are verified.

/ak:scenario "User registration with OAuth providers" --format test-scenarios

Modes / Routes

default

One-shot scenario table

Research:: Read target once
Review:: Security dimension when relevant
Validate:: Severity summary

--iterations N

Bounded iteration

Research:: Repeated novelty loop
Review:: Security focus possible
Validate:: Stops exactly at N iterations

--saturation

Saturation loop

Research:: Continues until novelty exhausted
Review:: Security focus possible
Validate:: Stops after two zero-novelty iterations

Sample Prompt

Concrete invocations for each flag, subcommand, mode, or route available in the live workflow.

One-shot API review Recommended

/ak:scenario src/api/payment.ts

Use when:: Use before implementing or testing a complex API path.
Expected:: A Scenario Report that filters the 12 dimensions, lists skipped assumptions, groups 3–5 scenarios per relevant dimension, and summarizes severity totals.

Bounded iteration

/ak:scenario src/api/payment.ts --iterations 25

Use when:: Use when you want exhaustive exploration but need an exact stop count.
Expected:: An iterative run that keeps new or variant situations, discards duplicates with reasons, prints progress every 5 iterations, and stops after exactly 25 iterations.

Security saturation

/ak:scenario src/middleware/auth.ts --saturation --domain security --focus security

Use when:: Use for a deep pre-release coverage audit where novelty should decide the stop point.
Expected:: A saturation loop that prioritizes security-relevant dimensions, logs kept and discarded cases to `scenario-results.tsv`, then halts after two consecutive zero-novelty iterations.

Test scenario output

/ak:scenario "User registration with OAuth providers" --format test-scenarios

Use when:: Use before writing tests when the immediate consumer is a QA or regression plan.
Expected:: Scenario rows shaped for test planning, including actor or input context, preconditions, expected behavior, severity, and the relevant decomposition dimension.

Handled Scope

Read target
Filter dimensions
Generate scenarios
Classify severity
Iterate when requested
Track novelty